Parallax

Macro-to-Onchain Correlation Research Agent on Mantle

An ERC-8004 reputation agent that quantifies how macro shocks (FOMC, CPI, IPO) propagate through Mantle's CeDeFi layer — SPCXx on Merchant Moe DLMM and Aave V3 stablecoin positioning, with verifiable on-chain evidence, not invented numbers.

Problem Statement

Tokenized RWAs are landing on Mantle — SPCXx (SpaceX exposure via xStocks), Fluxion liquidity, InsightX prediction markets, Aave V3 capital efficiency. Issuance is solved; distribution intelligence is not.

Researchers and traders still face a structural gap:

Pain	Why it matters
Macro ↔ on-chain disconnect	A CPI print or FOMC hold moves TradFi instantly, but Mantle's DLMM pools and Aave utilization shifts are scattered across RPC, indexers, and narrative Twitter threads.
No reproducible correlation layer	Analysts eyeball charts. There is no agent that snapshots pre/post event windows, computes surprise magnitude, and outputs Pearson cross-event statistics with provenance.
Trust deficit in AI research	LLM agents hallucinate price moves when chain data is missing. Judges and users cannot tell fiction from RPC.
No economic loop	One-off dashboards do not capture payment, consumption, validation, and reputation — the primitives Mantle ships (x402, ERC-8004) stay disconnected from research UX.

The harder challenge after issuance is distribution: moving from "asset exists onchain" to "we can measure and trust how capital repositioned when macro fired."

Solution

Parallax is a quantitative research terminal that closes the loop between macro catalysts and Mantle on-chain reaction — with a paywall, on-chain reputation, and anti-hallucination data quality gates.

What Parallax does

1. Stage macro events — 7 pre-built calendar events (FOMC, CPI, IPO) or dynamic discovery via Gemini web search (CPI April 2026, etc.).
2. Acquire three quantitative legs at historical Mantle mainnet blocks:

• Leg 1 — Macro surprise: FRED DFF / CPIAUCSL YoY proxy, or hardcoded IPO pricing surprise.
• Leg 2 — xStocks: SPCXx/USDT0 Merchant Moe LB getActiveId() price delta via historical RPC.
• Leg 3 — Aave V3: USDC / USDT0 / USDe getReserveData + getAssetPrice utilization, borrow APY, and supply USD deltas.

3. Correlate — Pearson r across events; Gemini synthesizes institutional-grade dossiers with ASCII quantitative charts.
4. Gate full reports behind x402 micropayments on Mantle Sepolia.
5. Validate — users grade Alpha via ERC-8004 on-chain feedback with IPFS-pinned consensus notes; reputation curve updates live.

The Tokenomic Flywheel

Parallax isn't just a dashboard; it's a self-sustaining ecosystem powered by Mantle primitives:

1. Paywall (x402): Users pay a nominal $MNT fee via x402 to unlock high-alpha institutional reports.
2. Synthesis: Users consume the Gemini-synthesized insights backed by ASCII quantitative charts.
3. Validation (ERC-8004): Users sign an on-chain transaction on Mantle Sepolia to grade the Alpha (e.g., 5/5) and attach an IPFS-pinned consensus note.
4. Reputation: The Agent's ERC-8004 on-chain reputation curve updates dynamically, increasing trust for future reports.

Anti-Hallucination by Design

• Deterministic On-Chain Grounding: Parallax eradicates LLM market hallucinations by entirely decoupling data acquisition from the synthesis layer. Every quantitative ledger item is deterministically extracted via live and historical Mantle RPC queries rather than generated by raw probabilistic inference.
• Graceful Infrastructure Gaps: When backtesting across older block windows where target asset pools were not yet deployed or third-party lending operations revert, the engine seamlessly traps the contract failures into a structural data_quality.caveats array and forces explicit null fields instead of fabricating filler metrics.
• Mathematical Integrity: Advanced statistical evaluations, including cross-layer Pearson r correlation matrices, strictly resolve to n/a when mathematically undefined (such as instances of zero variance or insufficient data pairs), refusing to bend mathematical reality for narrative convenience.
• Auditable Data Lineage: Every generated dossier embeds an immutable data-quality footprint tracking specific indexing profiles (xstocks_source, aave_source). This gives institutional users and verifiers absolute visibility into whether data was derived from historical RPC snapshots or live market states.

Tech Stack

Core

Layer	Technology
Frontend	Next.js 15, React
Backend	Node.js, TypeScript
Macro data	FRED API (DFF, CPIAUCSL)
On-chain reads	Mantle mainnet historical RPC — block-resolved eth_call at pre/post event windows
Agent identity	agent0-sdk, ERC-8004 on Mantle Sepolia (5003:308)
Payments	x402 v2 — QuestFlow facilitator, USDC EIP-3009 on Mantle Sepolia
Storage	IPFS via Pinata (ERC-8004 feedback URIs)

Mantle Ecosystem Integration

Mantle primitive	Parallax usage
Mantle AI Agent Skills (@mantleio/mantle-core)	In-process MCP tools via mantle-client.ts: mantle_resolveAddress, mantle_queryIndexerSql, mantle_getLendingMarkets, pool discovery — powers preflight, xStocks, and Aave acquisition
SPCXx / xStocks	Merchant Moe Liquidity Book pair 0xd14B0DcD… — historical getActiveId() price at event blocks
Merchant Moe DLMM	Concentrated liquidity price discovery for synthetic RWA exposure
Aave V3 on Mantle	Stablecoin utilization / borrow / supply snapshots at resolved historical blocks
Fluxion	Protocol config in mantle-core; SQL indexer hook for volume (optional)
ERC-8004	Agent 5003:308 registered on Mantle Sepolia; Identity + Reputation registries; readAllFeedback + IPFS comments
x402	GET /api/report/:eventId → HTTP 402 teaser + PAYMENT-REQUIRED; unlock full dossier on settlement
Mantle Sepolia	ERC-8004 feedback txs, x402 pay-to, wallet demo (chain 5003)
Mantle Mainnet	Historical RPC for correlation engine (Aave, Merchant Moe)

Mantle AI Agent Skills: The correlation pipeline is not a wrapper around a single API. backend/src/mcp/mantle-client.ts embeds mantle-core tools directly into acquisition, preflight, and baseline extraction. The same skill surface Mantle documents for research agents.

How It Works

Technical Architecture

Follow the data

Parallax is built for evidence-first research, not narrative-first blogging:

• Macro surprise is computed from FRED observations with explicit surprise_method (fred_dff, fred_cpi_yoy, hardcoded for IPO).
• On-chain deltas are block-anchored: resolveBlockAtTimestamp maps event windows to Mantle mainnet heights, then readContract at those heights.
• Cross-event Pearson matrix compares surprise magnitude vs. xStocks price Δ vs. Aave utilization Δ across the Q1–Q2 2026 calendar (7 events + dynamic discoveries).
• Computed Metrics in every dossier (macroSurprise, aaveUtilizationDelta, aaveSupplyDeltaUsd, pearsonMacroAave) trace directly to eventDelta — the LLM narrates, but numbers come from the engine.

Tell the story

Mantle is positioning as the distribution layer for RWAs — SPCXx brings SpaceX exposure onchain; Aave brings capital efficiency. Parallax asks: when CPI surprises to the upside, does stablecoin liquidity on Mantle tighten before xStocks react? The agent encodes that question as reproducible windows, not a one-off thread.

Track the narratives

Narrative (2026)	Parallax lens
Tokenized equities on Mantle (SPCXx, Fluxion, Bybit xStocks)	Leg 2 — Merchant Moe LB price / liquidity delta around macro windows
CeDeFi capital efficiency (Aave V3)	Leg 3 — utilization & supply flow direction (risk_on / risk_off / neutral)
AI agents for on-chain research	Mantle Skills + Gemini + ERC-8004 reputation compounding
Agent payments (x402)	Report paywall — micropayment before institutional synthesis

Map the landscape

Approach	Parallax differentiation
Generic LLM crypto chat	Block-anchored RPC + FRED; null when data missing
Static dashboards	Event-window correlation with Pearson cross-event stats
Unpaid research bots	x402 flywheel + ERC-8004 validated reputation
Off-chain reputation	On-chain readAllFeedback + IPFS consensus notes on Mantle Sepolia

Demo

https://drive.google.com/file/d/109i0Z6k2bmjTm_q1CAjdKjQv3VXTOFaD/view?usp=sharing

Future Roadmap

• Production x402 QuestFlow Integration: The underlying payment-gating architecture for the research agent is fully implemented. Currently pending approval for the submitted QuestFlow Facilitator API key, at which point the system will seamlessly transition from the current developer bypass into a fully live production monetization gateway.
• Mobile Viewport Optimization: Transition the current desktop-first, high-density quantitative HUD console into a fully responsive, mobile-optimized interface for on-the-go macro-to-onchain alpha validation.